PROJECT SUMMARY The cerebellum plays a key role in motor control, particularly in motor learning. More recently, the cerebellum has also been implicated in cognitive processing. Indeed, cerebellar damage is associated with a wide range of neurological and neuropsychatric disorders, including ataxia, dystonia, schizophrenia and autism. Despite this apparent functional heterogeneity, the cerebellar microcircuit is remarkably homogeneous, both across its different regions and across different animal species. Thus, it has been suggested that the cerebellum may accomplish its job by performing one universal computation, and then sending out the results of the computation to other brain areas, both motor and non-motor. Previous work indicates that the universal computation performed by the cerebellum involves using past experience to predict future events, including those that are caused by our own movements. The long-term objective of this project is to achieve a full mechanistic understanding of how the cerebellum learns to make these predictions. The focus will be on the error signals that are critical for alerting the cerebellum that a prediction was wrong and needs to be updated. In the three aims of this proposal, we examine: 1) positive prediction errors (when something unexpected happens), 2) negative prediction errors (when something expected doesn't happen), and 3) temporal- difference prediction errors (when a stimulus predicts that something is about to happen). All experiments are done on a newly developed treadmill apparatus for eyeblink conditioning in head-fixed mice. Eyeblink conditioning was chosen as the model system because it offers a number of advantages for the project: 1) The behaviorally-relevant error signals are under experimental control and can be easily manipulated, 2) The basic conditioning task can be modified to ask questions about the role of error signals in driving both motor learning and higher-order associations, and 3) The olivo-cerebellar regions that are critical for processing error signals have been identified. By combining the elegant simplicity of eyeblink conditioning with new technologies for optogenetics, electrophysiology, and two-photon calcium imaging, the proposed experiments will record and manipulate the error-related neural signals present during the learning process with an unprecedented level of temporal and cellular specificity. This research could help develop new therapeutic approaches to treat motor and cognitive disorders associated with cerebellar dysfunction, not by targeting molecular mechanisms of neural plasticity, but the instructive error-related signals that drive them. In this regard, the specific aims of the application are designed to ask not only ?what is the neural code for error signals in the cerebellum?, but also ?how can we manipulate the code to enhance learning??.